[PATCH 4.10 90/93] libnvdimm: fix nvdimm_bus_lock() vs device_lock() ordering

From: Greg Kroah-Hartman
Date: Thu May 18 2017 - 07:34:16 EST


4.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dan Williams <dan.j.williams@xxxxxxxxx>

commit 452bae0aede774f87bf56c28b6dd50b72c78986c upstream.

A debug patch to turn the standard device_lock() into something that
lockdep can analyze yielded the following:

======================================================
[ INFO: possible circular locking dependency detected ]
4.11.0-rc4+ #106 Tainted: G O
-------------------------------------------------------
lt-libndctl/1898 is trying to acquire lock:
(&dev->nvdimm_mutex/3){+.+.+.}, at: [<ffffffffc023c948>] nd_attach_ndns+0x178/0x1b0 [libnvdimm]

but task is already holding lock:
(&nvdimm_bus->reconfig_mutex){+.+.+.}, at: [<ffffffffc022e0b1>] nvdimm_bus_lock+0x21/0x30 [libnvdimm]

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #1 (&nvdimm_bus->reconfig_mutex){+.+.+.}:
lock_acquire+0xf6/0x1f0
__mutex_lock+0x88/0x980
mutex_lock_nested+0x1b/0x20
nvdimm_bus_lock+0x21/0x30 [libnvdimm]
nvdimm_namespace_capacity+0x1b/0x40 [libnvdimm]
nvdimm_namespace_common_probe+0x230/0x510 [libnvdimm]
nd_pmem_probe+0x14/0x180 [nd_pmem]
nvdimm_bus_probe+0xa9/0x260 [libnvdimm]

-> #0 (&dev->nvdimm_mutex/3){+.+.+.}:
__lock_acquire+0x1107/0x1280
lock_acquire+0xf6/0x1f0
__mutex_lock+0x88/0x980
mutex_lock_nested+0x1b/0x20
nd_attach_ndns+0x178/0x1b0 [libnvdimm]
nd_namespace_store+0x308/0x3c0 [libnvdimm]
namespace_store+0x87/0x220 [libnvdimm]

In this case '&dev->nvdimm_mutex/3' mirrors '&dev->mutex'.

Fix this by replacing the use of device_lock() with nvdimm_bus_lock() to protect
nd_{attach,detach}_ndns() operations.

Fixes: 8c2f7e8658df ("libnvdimm: infrastructure for btt devices")
Reported-by: Yi Zhang <yizhan@xxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>

---
drivers/nvdimm/btt_devs.c | 2 +-
drivers/nvdimm/claim.c | 23 +++++++++++++++--------
drivers/nvdimm/dax_devs.c | 2 +-
drivers/nvdimm/pfn_devs.c | 2 +-
4 files changed, 18 insertions(+), 11 deletions(-)

--- a/drivers/nvdimm/btt_devs.c
+++ b/drivers/nvdimm/btt_devs.c
@@ -314,7 +314,7 @@ int nd_btt_probe(struct device *dev, str
if (rc < 0) {
struct nd_btt *nd_btt = to_nd_btt(btt_dev);

- __nd_detach_ndns(btt_dev, &nd_btt->ndns);
+ nd_detach_ndns(btt_dev, &nd_btt->ndns);
put_device(btt_dev);
}

--- a/drivers/nvdimm/claim.c
+++ b/drivers/nvdimm/claim.c
@@ -21,8 +21,13 @@
void __nd_detach_ndns(struct device *dev, struct nd_namespace_common **_ndns)
{
struct nd_namespace_common *ndns = *_ndns;
+ struct nvdimm_bus *nvdimm_bus;

- lockdep_assert_held(&ndns->dev.mutex);
+ if (!ndns)
+ return;
+
+ nvdimm_bus = walk_to_nvdimm_bus(&ndns->dev);
+ lockdep_assert_held(&nvdimm_bus->reconfig_mutex);
dev_WARN_ONCE(dev, ndns->claim != dev, "%s: invalid claim\n", __func__);
ndns->claim = NULL;
*_ndns = NULL;
@@ -37,18 +42,20 @@ void nd_detach_ndns(struct device *dev,
if (!ndns)
return;
get_device(&ndns->dev);
- device_lock(&ndns->dev);
+ nvdimm_bus_lock(&ndns->dev);
__nd_detach_ndns(dev, _ndns);
- device_unlock(&ndns->dev);
+ nvdimm_bus_unlock(&ndns->dev);
put_device(&ndns->dev);
}

bool __nd_attach_ndns(struct device *dev, struct nd_namespace_common *attach,
struct nd_namespace_common **_ndns)
{
+ struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(&attach->dev);
+
if (attach->claim)
return false;
- lockdep_assert_held(&attach->dev.mutex);
+ lockdep_assert_held(&nvdimm_bus->reconfig_mutex);
dev_WARN_ONCE(dev, *_ndns, "%s: invalid claim\n", __func__);
attach->claim = dev;
*_ndns = attach;
@@ -61,9 +68,9 @@ bool nd_attach_ndns(struct device *dev,
{
bool claimed;

- device_lock(&attach->dev);
+ nvdimm_bus_lock(&attach->dev);
claimed = __nd_attach_ndns(dev, attach, _ndns);
- device_unlock(&attach->dev);
+ nvdimm_bus_unlock(&attach->dev);
return claimed;
}

@@ -114,7 +121,7 @@ static void nd_detach_and_reset(struct d
struct nd_namespace_common **_ndns)
{
/* detach the namespace and destroy / reset the device */
- nd_detach_ndns(dev, _ndns);
+ __nd_detach_ndns(dev, _ndns);
if (is_idle(dev, *_ndns)) {
nd_device_unregister(dev, ND_ASYNC);
} else if (is_nd_btt(dev)) {
@@ -184,7 +191,7 @@ ssize_t nd_namespace_store(struct device
}

WARN_ON_ONCE(!is_nvdimm_bus_locked(dev));
- if (!nd_attach_ndns(dev, ndns, _ndns)) {
+ if (!__nd_attach_ndns(dev, ndns, _ndns)) {
dev_dbg(dev, "%s already claimed\n",
dev_name(&ndns->dev));
len = -EBUSY;
--- a/drivers/nvdimm/dax_devs.c
+++ b/drivers/nvdimm/dax_devs.c
@@ -124,7 +124,7 @@ int nd_dax_probe(struct device *dev, str
dev_dbg(dev, "%s: dax: %s\n", __func__,
rc == 0 ? dev_name(dax_dev) : "<none>");
if (rc < 0) {
- __nd_detach_ndns(dax_dev, &nd_pfn->ndns);
+ nd_detach_ndns(dax_dev, &nd_pfn->ndns);
put_device(dax_dev);
} else
__nd_device_register(dax_dev);
--- a/drivers/nvdimm/pfn_devs.c
+++ b/drivers/nvdimm/pfn_devs.c
@@ -484,7 +484,7 @@ int nd_pfn_probe(struct device *dev, str
dev_dbg(dev, "%s: pfn: %s\n", __func__,
rc == 0 ? dev_name(pfn_dev) : "<none>");
if (rc < 0) {
- __nd_detach_ndns(pfn_dev, &nd_pfn->ndns);
+ nd_detach_ndns(pfn_dev, &nd_pfn->ndns);
put_device(pfn_dev);
} else
__nd_device_register(pfn_dev);